MULTI-OBJECTIVE OPTIMIZATION OF A SIMPLIFIED CAR BODY USING COMPUTATIONAL FLUID DYNAMICS

Today's strict fuel economy requirement produces the need for the cars to have really optimized shapes among other characteristics as optimized cooling packages, reduced weight, to name a few. With the advances in automotive technology, tight global oil resources, lightweight automotive design process becomes a problem deserving important consideration. It is not however always clear how to modify the shape of the exterior of a car in order to minimize its aerodynamic resistance. Air motion is complex and operates differently at different weather conditions. This gap can be covered by the use of adjoint solvers which predict the sensibility of the aerodynamic forces to changes of the geometry. Alternatively, Computational Fluid Dynamics (CFD) solvers can be partnered with optimization software which guide model design changes and evaluate the corresponding results. Design changes can be executed by modifying a parameterized geometry or using mesh morphing techniques. With the advances in computational fluid dynamics, design optimization methods in the aerodynamic design are more important than ever. In the present paper, ANSYS Fluent will be used in conjunction with the optimization software ANSYS DesigmXplorer to study ways of reducing drag and lift for a simplified car body. ANSYS simulation software allows one to predict, with confidence, the impact of fluid flows on the product throughout design and manufacturing as well as during end use. CFD is a complex technology involving strongly coupled non-linear partial differential equations which attempt to computationally simulate theoretical and experimental models in a discrete domain of complex geometric shape. A detailed assessment of errors and uncertainties has to concern itself with the three roots of CFD: theory, experiment, and computation. Further, the application of CFD is rapidly expanding with the growth in computational resources. The body in question in this study is the Ahmed body [1] which has been used numerous times for CFD code validation. This geometry represents a road legal car which is used to study the effect of different forces like, aerodynamic drag force, lift force, and some other major forces which affect a car's motion significantly. Despite being a simple body, accurate prediction of its aerodynamic performance often requires very accurate and computationally expensive calculations. We would like to investigate if meaningful optimizations can be achieved by using reduced resources, by analyzing how air at different velocity affect the body and what changes might be necessary for a further optimized performance. The purpose here is not to predict the absolute values of drag for this body, but to demonstrate that optimization can be performed with limited resources relying on information about drag deltas rather than absolute values. Keeping limiting resources in mind, a grid independence study wasn't done.